Role of Coronary Angiography in Pre–Liver Transplantation Cardiac Evaluation: Experience From an Asian Transplant Institution

Role of Coronary Angiography in PreeLiver Transplantation Cardiac Evaluation: Experience From an Asian Transplant Institution N.Q. Panga, W.C.A. Kowa,*, J.H. Lawb, L.T.T. Panc, B.L.K. Limd, C.C.R. Wonge, K.Y.S. Changa, I.S. Ganpathia, and K. Madhavana a Department of Surgery, National University Health System, Singapore; bYong Loo Lin School of Medicine, National University of Singapore, Singapore; cDepartment of Anaesthesiology, National University Health System, Singapore; dDepartment of Gastroenterology and Hepatology, National University Health System, Singapore; and eDepartment of Cardiology, National University Health System, Singapore

ABSTRACT Background. Liver transplant (LT) patients with significant coronary artery disease (CAD) have poorer outcomes. Pre-LT coronary angiography (CA) is associated with significant complications in cirrhotic patients. Methods. This study aimed to identify predictors of abnormal CA in pre-LT cardiac assessment and to develop a predictive model to reduce unnecessary CA. From January 2006 to June 2013, 122 patients underwent CA based on the current institutional protocol. Results. Forty-one (33.6%) patients had abnormal CA. Univariate analysis showed age
65 years (P ¼ .001), cryptogenic cirrhosis (P ¼ .046), cardiac comorbidities (P ¼ .027), ischemic heart disease (IHD; P ¼ .002), left ventricular hypertrophy (LVH; P ¼ .004), hypertension (P ¼ .002), diabetes mellitus (P ¼ .017), dyslipidemia (P < .001), metabolic syndrome (P ¼ .003),
2 CAD risk factors (P ¼ .001), and high Framingham risk score (hard CAD risk, P ¼ .018; cardiovascular disease: lipids, P ¼ .002; body mass index, P < .001) to be significant predictors of abnormal CA. A predictive model was developed with the use of multivariable logistic regression and included diabetes, dyslipidemia, IHD, age
65 years, and LVH, achieving a specificity of 55.1% and sensitivity of 90.0%. This would reduce unnecessary CA by up to one-half in our study population (from 81 to 35) while maintaining a false negative rate of only 8.5%. Conclusions. Diabetes, dyslipidemia, IHD, age
65 years, and LVH appear to be predictors of abnormal CA in pre-LT patients. Our predictive model may help to better select patients for CA, although further validation is required.

T

HE PREVALENCE of coronary artery disease (CAD) in end-stage liver disease (ESLD) patients has been estimated to be 2.5%e27% [1e3]. Tiukinhoy-Laing et al [3] studied liver transplant (LT) candidates with no history of CAD evaluated by means of cardiac catheterization and found that 20% had severe CAD, 26% had moderate CAD, 36% had mild CAD, and 18% had no CAD. Indeed, LT is a high-risk operative procedure, and the presence of CAD in these patients is associated with poorer outcomes, with increased mortality and morbidity rates of 50% and 81%, respectively [4e6]. Yet current guidelines for pre-LT cardiac assessment are conflicting and the protocol is different in individual ª 2017 Elsevier Inc. All rights reserved. 230 Park Avenue, New York, NY 10169

Transplantation Proceedings, 49, 1797e1805 (2017)

institutions [7]. The 2005 American Association for the Study of Liver Diseases practice guidelines recommended that patients over the age of 50 years or those with cardiac risk factors (clinical or family history of heart disease, diabetes mellitus [DM], chronic smoker) should undergo coronary angiography (CA) [8]. In 2009, the European Society

*Address correspondence to Assistant Professor Wei Chieh Alfred Kow, Division of Hepatobiliary and Pancreatic Surgery and Liver Transplantation, University Surgical Cluster, National University Hospital, 1E, Kent Ridge Road, NUHS Tower Block, Level 8, Singapore 119228. E-mail: [email protected] 0041-1345/17 http://dx.doi.org/10.1016/j.transproceed.2017.04.021

1797

1798

of Cardiology published their guidelines for perioperative cardiac risk assessment and management in noncardiac surgeries [9]: LT surgery was classified as intermediate cardiac risk, and preoperative CA was recommended. In some of the more recent publications on this controversy, in 2011 Raval et al published a state-of-the-art paper recommending that patients with
2 traditional cardiac risk factors should undergo CA before LT [10]. On the other hand, the American Heart Association and American College of Cardiology Foundation published their guidelines in 2007 for perioperative cardiac evaluation in noncardiac surgery, and an expert consensus document in 2012 stated that noninvasive cardiac stress tests should be considered in patients with
3 cardiac risk factors [11,12]. There were no recommendations on the use of CA as a preoperative screening modality in either of those publications. Currently, CA is the best modality for diagnosing CAD. Indeed, coronary revascularization during CA could reduce the perioperative risk for patients undergoing LT with left main disease or severe proximal 3-vessel disease, especially in the presence of reduced left ventricular systolic function [13,14]. However, not all patients who are referred for LT workup need to undergo CA, and a selective approach would be more desirable. Therefore, the present study was undertaken with the primary aim of identifying predictors of abnormal CA in pre-LT patients. The secondary aim was to develop a predictive model to better select patients who will benefit from CA. Careful selection of pre-LT candidates could reduce unnecessary CA in this group of patients.

METHODS There were 391 patients referred to the National University Center for Organ Transplantation (NUCOT) at the National University Hospital (NUH) for LT consideration from January 2006 to June 2013. In this case-control study, 146 patients who were deemed to be potentially suitable candidates for LT and who completed the full cardiac evaluation were retrospectively reviewed, of which 122 underwent CA based on the institutional protocol. Most of the pre-LT workup assessments were done on an inpatient basis. In addition to the regular blood investigations, endoscopic assessment, and imaging scans, cardiac evaluation is an integral part of the assessment. Based on the current institutional protocol, all potential LT patients underwent electrocardiography and 2-dimensional (2D) echocardiography as baseline cardiac investigations. In addition, patients who were
50 years of age or had cardiovascular risk factors (chronic smoker, history of heart disease, DM, hyperlipidemia, and hypertension) were subjected to CA. On the other hand, patients who were <50 years of age and with no cardiovascular risk factors underwent noninvasive stress testing, which could be either dobutamine stress echo (DSE) or myocardial perfusion scan (MPS). If found to be abnormal, they subsequently underwent CA and treatment could be instituted in the same setting should any significant CAD be detected. Before CA, the platelet count and coagulation profile were corrected appropriately. In cirrhotic patients needing LT, thrombocytopenia is very common and multiple platelet transfusions may be required before the procedure. This can prolong hospitalization

PANG, KOW, LAW ET AL during the work-up process. Likewise, the prolonged coagulation profile requires transfusion of plasma products accordingly. Other systems that may require optimization in pre-LT patients include renal function. Patients were scheduled for CA only after optimal preparation was achieved. Cardiologists in our institution routinely perform angiography via the radial artery approach. Local anesthetic agents were provided before access to radial artery was created. If significant CAD was detected, the cardiologists would communicate with the LT team before deciding on the type of coronary intervention, including the type of stents to use and the duration and types of antiplatelet agents to be administered. Patient demographics, intraoperative details, perioperative outcomes, and details of their cardiac work-up were gathered by reviewing electronic medical records and patients’ case notes. Complete findings of CAs performed were also clearly documented and categorized. This study was approved by the Institutional Review Board of the National Healthcare Group (NHG DSRB ref 2013/00851). Diabetes was defined as pre-transplantation fasting glucose >7.0 mmol/L or if the patient was on an oral hypoglycemic agent or insulin. Pre-transplantation dyslipidemia was defined as triglyceride level
200 mg/dL or high-density lipoprotein (HDL) <40 mg/dL or if the patient was on a lipid-lowering drug. Pre-transplantation hypertension was defined as
2 blood pressure readings >140/90 mm Hg or if the patient was on an antihypertensive medication. Metabolic syndrome was defined as meeting
3 of the following: Body mass index (BMI) > 27.5 kg/m2 Triglyceride
150 mg/dL (1.7 mmol/L) or on pharmacologic treatment HDL <40 mg/dL (1.0 mmol/L) for men or <50 mg/dL (1.3 mmol/L) for women, or on pharmacologic treatment Blood pressure
130/85 mm Hg or on pharmacologic treatment Fasting glucose
110 mg/dL (6.1 mmol/L) or on pharmacologic treatment This definition is based on the National Cholesterol Education Program’s Adult Treatment Panel III report (ATPIII) [15]. Although the report uses waist circumference as one of the criteria, we used BMI as a surrogate measure because information on waist circumference was not available in this retrospective review. The Framingham risk score (FRS) is a sex-specific algorithm used to estimate the 10-year cardiovascular risk of an individual (CAD, cerebrovascular events, peripheral arterial disease, heart failure, myocardial infarction, or coronary death) [16]. The FRS was calculated for all patients with the use of 3 different calculators (1 each using lipids and BMI and the 3rd predicting hard coronary heart disease) developed by the Framingham Heart Study Group. Patients were classified to have abnormal CA if any stenosis of any named coronary vessels were noted. Coronary arteries that were evaluated included the left anterior descending, left circumflex, and right coronary and their major branches. In adopting “any stenosis” as a positive CA finding, this study tried to include as many patients as possible as having CAD. This is an acknowledgement of the fact that any stenosis in named coronary arteries can lead to significant morbidity. This also gives rise to a smaller group of patients who are classified as having normal coronary circulation. Despite this, a significant number of patients with normal coronary circulation are still being subjected to CA based on current selection criteria. If patients with 10%e50% stenosis in the named coronary vessels are classified as having negative CA (ie, “normal” CA), the number of patients who undergo unnecessary CAs would be even larger. Results were analyzed with the use of SPSS version 21.0, and P < .05 was considered to be statistically significant. Univariate

CORONARY ANGIOGRAPHY BEFORE LIVER TRANSPLANT analysis was performed with the use of Mann-Whitney U test for continuous variables and Fisher exact test for categoric variables. A logistic regression model was used to identify possible predictors of abnormal CA.

RESULTS

Of the 146 patients who were referred and underwent a complete cardiac evaluation, 122 underwent CA as part of their transplantation work-up based on current institutional protocol. None of those patients had any CA-related complications. Although there were no complications of CA in our series, 22 patients (18.0%) required transfusion of blood products before CA to correct coagulopathy and/or thrombocytopenia to an acceptable international normalized ratio (INR) of 1.5 and platelet count of 50  109/L. Among the 22 patients who required transfusions before CA, 9 (40.9%), 12 (54.5%), and 15 (68.2%) received packed cell, fresh frozen plasma, and platelet transfusions, respectively, and each patient received a median of 4.5 units (range, 1.0e14.0 units) of blood products. Although no complications occurred during the transfusion of blood products, patients were exposed to the potential risks of transfusions, including antigen-antibody reaction, hemolysis, fluid overload, and infections. Moreover, patients who were admitted during the LT work-up that included the CA were required to be hospitalized for a median of 3 additional days (range, 2e4 days) for the correction of coagulopathy/thrombocytopenia before CA and for postprocedure monitoring. These factors translate to increased costs and duration of the pre-transplantation work-up. Table 1 presents the general profile of our study population. The median age of the study population was 55 (range, 23e72) years. Nearly three-fourths were male (n ¼ 109; 74.7%), and the most common etiology of chronic liver disease was hepatitis B cirrhosis (n ¼ 54; 37.0%). Nearly one-half of the study population (n ¼ 69; 47.3%) subsequently underwent LT. Figure 1 shows the distribution of the study population. Among the 122 patients who underwent CA, 81 (66.4%) turned out to have a normal CA and slightly more than onethird (n ¼ 41; 33.6%) had abnormal CAs with varying degrees of severity of CAD. A large majority of the patients with normal CA (n ¼ 77, 95.1%) were subsequently deemed to be fit for LT. Of the 4 patients with normal CA who were not fit for transplantation, 2 were rejected owing to psychiatric conditions, 1 owing to severe pulmonary hypertension, and 1 owing to severe heart failure. Among the 41 patients with abnormal CA, 30 (73.2%) were deemed to be fit to proceed with LT. There were 11 patients with abnormal CA who were deemed to be unsuitable for LT. One was rejected owing to progression of hepatocellular carcinoma outside of the transplantation criteria, and another was rejected owing to severe aortic stenosis. The remaining 9 had severe CAD (Table 2). Of the 30 patients with abnormal CA who subsequently underwent LT, none of them required coronary intervention before the operation.

1799

Predictors of Abnormal Coronary Angiography

Table 1 presents an analysis of patient characteristics and disease factors between patients with normal and abnormal CAs. Patients with abnormal CA were 8 times more likely to be
65 years of age (19.5% vs 2.5%; P ¼ .001). Patients with a history of ischemic heart disease (IHD) were understandably more likely to have an abnormal CA (19.5%) compared with normal CA (2.5%; P ¼ .002). Interestingly, patients with abnormal CA (31.7%) were nearly twice as likely to have cryptogenic cirrhosis compared with patients with normal CA (16.0%; P ¼ .046). Detailed analysis of all patients’ 2D echocardiography findings showed that patients with abnormal CA were more likely to have left ventricular hypertrophy (LVH; 27.5% vs 7.7%; P ¼ .004; Table 3). In addition, a significantly higher proportion of patients with abnormal CA were found to have underlying hypertension (58.5% vs 29.6%; P ¼ .002), DM (58.5% vs 35.8%; P ¼ .017), and dyslipidemia (31.7% vs 6.2%; P < .001). However, BMI was not found to be significantly different between the 2 groups (abnormal CA: mean BMI, 25.6  4.6 kg/m2; normal CA: mean BMI, 24.8  5.2 kg/m2; P ¼ .400). The patient distributions of various weight categories were also similar between the 2 groups (Table 3). Despite that, there was a significantly higher prevalence of metabolic syndrome in patients with abnormal CA (56.1%) compared with patients with normal CA (28.4%; P ¼ .003). The same finding was noted when the FRS was calculated: all 3 scores were significant predictors of abnormal CA. Finally, a significantly higher proportion of patients with abnormal CA results had
2 risk factors (abnormal CA 87.8% vs normal CA 58.0%; P ¼ .001). Predictive Model of Abnormal Coronary Angiography

A multivariable logistic regression model was developed with the 8 factors that were statistically significant on univariate analysis. These factors included IHD, metabolic syndrome, DM, hypertension, dyslipidemia, age, presence of cryptogenic cirrhosis, and LVH. Stepwise backward elimination with the use of P
.06 as the removal criteria left 5 variables remaining in the final model. These included age, dyslipidemia, DM, IHD, and LVH. We used this model to predict patients who could potentially avoid CA, thereby decreasing unnecessary CA. The probability of a patient having an abnormal CA, P, can be calculated with the use of the following equation: Y ¼  2:074 þ 2:404 ðage
65 yÞ þ 1:578 ðdyslipidemiaÞ þ 1:117 ðDMÞ þ 1:810 ðIHDÞ þ 1:410 ðLVHÞ where Y ¼ ln(P/[1  P]) and, therefore, P ¼ eY/(1 þ eY). This model is well calibrated (Hosmer-Lemeshow goodnessof-fit test, P ¼ .621), and by plotting a receiver operating characteristic (ROC) curve, as shown in Fig 2, it was found that using a cutoff value of P ¼ .25 gave the model a specificity of 55.1% and sensitivity of 90.0%. This calibration of high sensitivity at the expense of specificity

1800

PANG, KOW, LAW ET AL

Table 1. Comparison of Patient Characteristics and Disease Factors Between Patients With Normal and With Abnormal Coronary Angiography (CA), n (%) Variable

n Age, y, median (range) Age
65 y Sex Male Female Ethnicity Chinese Malay Indian Other Underwent LT Fit for transplantation Patient comorbidities Cardiac comorbidities* History of IHD History of arrhythmias History of valvular heart disease History of congestive cardiac failure History of cerebrovascular events Renal comorbidities Diabetic nephropathy Hepatorenal syndrome Pulmonary comorbidities Diagnosis HBV flare HBV cirrhosis HCV cirrhosis Alcoholic cirrhosis Cyptogenic cirrhosis Autoimmune hepatitis Primary biliary cirrhosis Drug-induced liver disease Wilson disease Others Cryptogenic cirrhosis Noncryptogenic cirrhosis Hepatocellular carcinoma Acute/fulminant hepatic failure MELD score, median (range)

General Profile

122 55 (23e72)

Abnormal CA

Normal CA

41 (33.6)

81 (66.4)

8 (19.5)

2 (2.5)

109 (74.7) 37 (25.3)

34 (82.9) 7 (17.1)

58 (71.6) 23 (28.4)

96 9 11 30 69 106

(65.8) (6.2) (7.5) (20.5) (47.3) (86.9)

27 2 2 10 11 30

(65.9) (4.9) (4.9) (24.3) (26.8) (73.2)

55 6 8 12 40 77

(67.9) (7.4) (9.9) (14.8) (49.4) (95.1)

20 10 4 1 4 3 27 5 9 19

(16.4) (8.2) (3.3) (0.8) (3.3) (2.5) (22.1) (4.1) (7.4) (15.6)

11 8 1 1 0 1 8 2 1 8

(26.8) (19.5) (2.4) (2.4) (0.0) (2.4) (19.5) (4.9) (2.4) (19.5)

9 2 3 0 4 2 19 3 8 11

(11.1) (2.5) (3.7) (0.0) (4.9) (2.5) (23.5) (3.7) (9.9) (13.6)

5 54 18 22 29 3 4 3 3 3 26 96 58 4 16.3

(3.4) (37) (12.3) (15.1) (19.9) (2.1) (2.7) (2.1) (2.1) (2.1) (21.3) (78.7) (47.5) (2.7) (6e45)

4 15 3 5 13 0 0 1 0 0 13 28 18 3 16.9

(9.8) (36.6) (7.3) (12.2) (31.7) (0.0) (0.0) (2.4) (0.0) (0.0) (31.7) (68.3) (43.9) (7.3) (7e34)

0 34 13 12 13 2 4 1 1 1 13 68 40 1 15.9

(0.0) (42.0) (16.0) (14.8) (16.0) (2.5) (4.9) (1.2) (1.2) (1.2) (16.0) (84.0) (49.4) (1.2) (6e45)

P Value

.001† .170

.523

.017† <.001† <.027† .002† .999 .336 .299 .999 .620 .999 .270 .393 .034†

.046† .567 .110 .461

Abbreviations: LT, liver transplantation; IHD, ischemic heart disease; HBV, hepatitis V virus; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease. *Cardiac comorbidities include ischemic heart disease, valvular heart disease, cardiac arrhythmias and congestive cardiac failure. † Significant P value (<.05).

is desired because although this model is primarily a tool to decrease unnecessarily CA, potential transplant patients with CAD should not be missed. Despite this restriction, the model gives a false positive rate (which represents unnecessary CA) of 49.3%, representing a significant reduction of unnecessary CA from 66.4% in our study population. The positive predictive value of the test is 50.7%, and the negative predictive value is 91.5%. This model was applied to the study population, and the results are shown in Fig 2. They showed that with this predictive model, the number of patients receiving CA that ultimately turns out to be normal could be reduced from 81 to 35, representing a decrease in unnecessary CA by more than one-half.

This model was further tested on the study population as presented in Table 2. A detailed analysis of the 9 patients with abnormal CA who were deemed to be unfit for LT owing to significant CAD was performed. Only 1 patient among those 9 had a calculated probability (of abnormal CA) of < .25 and would not have undergone a CA based on the predictive model. This was a 58-year-old man with no cardiovascular risk factors who turned out to have double vessel disease. DISCUSSION

Current literature has established the increased prevalence of cardiovascular risk factors in LT candidates and

CORONARY ANGIOGRAPHY BEFORE LIVER TRANSPLANT

1801

Fig 1. Distribution of the study population. Abbreviations: CA, coronary angiography; LT, liver transplantation.

setting. Most guidelines recommend routine 2D echocardiography as a baseline test, but its use as a sole screening modality is limited [20]. The use of other noninvasive investigations, eg, DSE, MPS, and newer modalities such as CT arteriography (CTA), has mixed verdicts. CTA has been shown to be effective in assessing both peri- and postoperative cardiovascular risk, potentially emerging as a replacement for CA [21,22]. The value of other modalities, however, remains undetermined. The stress tests are often

recipients [17] and the associated cardiovascular morbidity after surgery [18,19]. Therefore, accurate identification of patients who require CA before transplantation is an important step toward successful LT. Although various strategies for preoperative cardiac assessment of LT candidates have been proposed, there is currently no guideline that is universally adopted. Numerous studies have investigated the role of noninvasive cardiac tests to elucidate their role in the pre-LT

Table 2. Patients With Abnormal Coronary Angiography (CA) (Patients Rejected by LT Team Owing to Significant Coronary Artery Disease) Patient

1 2 3 4 5 6 7 8 9

Calculated Probability

Number of Positive Predictors

Age
65 y

IHD

DM

Dyslipidemia

LVH

1-, 2-, or 3-Vessel Disease (
50% Stenosis) on CA

0.650 0.112* 0.946 0.650 0.277 0.277 0.611 0.650 0.611

2 0 3 2 1 1 2 2 2

No No Yes No No No No No No

No No No No No No No No No

Yes No Yes Yes Yes Yes Yes Yes Yes

Yes No No Yes No No No Yes No

No No Yes No No No Yes No Yes

3 2 1 1 2 2 2 2 2

Abbreviations: DM, diabetes mellitus; LVH, left ventricular hypertrophy; others as in Table 1. *Calculated probability of <0.25 but with abnormal CA (patient would not have undergone CA based on the predictive model).

1802

PANG, KOW, LAW ET AL

Table 3. Comparison of 2-Dimensional Echocardiographic Characteristics and Cardiac Risk Factors Between Patients With Normal and With Abnormal CA Factor

Abnormal transthoracic echocardiography Dilated right ventricle Right ventricular hypertrophy Dilated left ventricle LVH Regional wall motion abnormality Impaired LVEF Aortic valve pathology Mitral valve pathology Tricuspid valve pathology Pulmonary valve pathology Pulmonary hypertension Prolonged left ventricular relaxation Cardiac disease detected on work-up Pulmonary hypertension Arrhythmias Moderate to severe valvular disease Heart failure Cardiac risk factors Known hypertension Systolic blood pressure, mm Hg Diastolic blood pressure, mm Hg Known DM Known insulin-dependent DM Known dyslipidemia High low-density lipids (
4.1) Low high-density lipids (<1.0) High total cholesterol (
6.2) High triglycerides (
2.3) BMI category Not overweight Overweight Obese Mean BMI, kg/m2 Weight, kg Metabolic syndrome Smoking Alcohol Framingham risk scores Lipids BMI Hard coronary heart disease Total number of CAD risk factors*
2 risk factors
1 risk factor

Abnormal CA

Normal CA

P Value

1 0 3 11 3 0 14 22 24 6 5 4

(2.5) (0.0) (7.5) (27.5) (7.5) (0.0) (35.0) (55.0) (60.0) (15.0) (12.5) (10.0)

5 1 8 6 5 2 27 46 52 9 12 8

(6.4) (1.3) (10.3) (7.7) (6.4) (2.6) (34.6) (59.0) (66.7) (11.5) (15.4) (10.3)

.662 .999 .748 .004† .999 .548 .967 .679 .474 .593 .673 .999

4 2 4 0

(9.8) (4.9) (9.8) (0.0)

10 2 5 4

(12.3) (2.5) (6.2) (4.9)

.771 .602 .483 .299

24 125 70 24 9 13 0 27 0 2

(58.5) (95e177) (52e98) (58.5) (22.0) (31.7) (0.0) (65.9) (0.0) (4.9)

24 120 70 29 13 5 2 50 2 4

(29.6) (89e171) (40e99) (35.8) (16.0) (6.2) (3.1) (64.1) (2.6) (5.1)

14 13 14 25.6 69.0 23 12 15

(34.1) (31.7) (34.1) (18.7e36.1) (46.0e98.5) (56.1) (29.3) (36.6)

29 27 25 24.8 67.0 23 21 28

(35.8) (33.3) (30.9) (15.9e36.0) (40.2e99.0) (28.4) (25.9) (34.6)

15.0 21.0 6.0 3 36 3

(1.1e89.4) (2.1e58.3) (0.1e29.0) (0e6) (87.80) (7.3)

11.3 11.4 4.0 2 47 12

(2.2e54.3) (2.4e48.9) (0.1e27.0) (0e6) (58.0) (14.8)

.002† .066 .638 .017† .423 <.001† .540 .849 .545 .999 .935

.400 .584 .003† .695 .826 .002† <.001† .018† <.001† .001† .234

Note. Values are presented as n (%) or median (range). Abbreviations: LVEF, left ventricular ejection fraction; BMI, body mass index; CAD, coronary artery disease; others as in Tables 1 and 2. *Coronary artery disease risk factors include hypertension, diabetes mellitus, dyslipidemia, smoking, obese/overweight, metabolic syndrome, personal history of ischemic heart disease, personal history of cerebrovascular accident, and peripheral vascular disease. † Significant P value (<.05).

limited by the poor exercise capacity of ESLD patients. Garg et al released a state-of-the-art paper in 2013 reviewing these noninvasive modalities [22]. They found that various reports have put the sensitivity and specificity of DSE to be 12.5%e100.0% and 57.1%e100.0%, respectively [2,23e25]. For single-photon emission computerized tomographic imaging, sensitivities of 36.8%e100.0% and specificities of 60.9%e62.5% were reported [9,26]. Such varied

results have inadvertently led to the proliferation of conflicting guidelines. With the uncertainty surrounding noninvasive cardiac tests, CA remains the criterion standard for the identification of CAD in the LT population [27,28]. It would be foolhardy, however, to prescribe a one-size-fits-all approach. In addition to the issue of additional cost for CA, the potential complications that this invasive procedure carries

CORONARY ANGIOGRAPHY BEFORE LIVER TRANSPLANT

1803

Fig 2. Receiver operating characteristic (ROC) curve for predictive model. Abbreviation: AUC, area under the ROC curve.

must be taken into consideration. Patients with ESLD are at increased risk of bleeding complications owing to underlying thrombocytopenia and coagulopathy [29]. Sharma et al reported that 88 ESLD patients undergoing CA had lower baseline hemoglobin and higher INR and serum creatinine levels than matched control subjects. Patients with ESLD had a higher rate of vascular complications such as pseudoaneuryms and higher rates of requirements for red cell, fresh frozen plasma, and platelet transfusions. In addition, major bleeding with angiography occurred in 14.8% of the ESLD group versus 3.7% of matched control group [30].

Optimization of patients’ coagulation profile, platelet levels, and renal functions for CA often prolongs the hospitalization for work-up. None of the 30 patients with abnormal CA who subsequently underwent LT in the present study required coronary intervention. This seems to indicate that a significant portion of the population with abnormal CA have only minor CAD and for them the risks of CA outweigh the benefits. The cardiovascular risk factors for abnormal CA that have been found in this study are similar to those described in the current literature, which were neatly summarized by

1804

Mandell, stating that age >50 years, male sex, DM, and obesity are the most common [31]. At the same time, Tiukinhoy-Laing et al found that patients with >1 cardiovascular risk factor (including male sex, age >50 years, smoking, hypercholesterolemia, and DM) have a greater probability of having significant CAD [3]. The demographics of the present study population were similar to those in Tiukinhoy-Laing et al’s study, with differences noted only in ethnicity (they had a predominantly [55%] white population). The majority (66%) of their study population was male, and viral hepatitis (51%) was the most common etiology for ESLD. Among patients with abnormal CA, 41.2% had hypertension (vs 58.5% in our study) and 44.3% had DM (vs 58.5% in our study). Unlike in our study, however, they did not find dyslipidemia to be a significant predictor of abnormal CA. To date, there has been no large-scale prospective study investigating the predictors of abnormal CA in the ESLD population. Most recommendations for cardiovascular risk factors set the age at 50 years. By using the ROC curve, however, it was found that a cutoff of 65 years gave the best specificity. No other studies have used a similar method of analysis to establish an age cutoff of 65 years as a risk factor for abnormal CA. Cryptogenic cirrhosis was also shown to be a predictive factor of abnormal CA in the present study. This unsurprising finding supports the proposal that the underlying etiology of cryptogenic cirrhosis may be related to burnt-out nonalcoholic steatohepatitis (NASH), which is both an inflammatory and atherogenic condition [32]. NASH results in an elevation of circulating inflammatory markers, even compared with patients with DM alone or other types of liver disease [33]. This can lead to burnt-out NASH presenting as cryptogenic cirrhosis. NASH has also been intimately associated with a significantly increased risk of CAD and its associated risk factors, such as hypertension, obesity, and DM [34,35]. Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease worldwide and has an estimated prevalence of up to 46% depending on geographic location [36]. This is important because NAFLD may lead to NASH, and NASH-associated factors, such as obesity, DM, and hypertension, may contribute to increased morbidity and mortality after LT [37]. Indeed, it should be expected that obesity is related to CAD and thus increases the likelihood of abnormal CA. Surprisingly, BMI was not observed to be a significant predictor of abnormal CA in our study. NUCOT has very stringent selection criteria for LT based on BMI, inadvertently introducing an element of selection bias. In addition, it must be noted that the BMI definition of the various classes of obesity is based on the ATPIII definition and its subsequent revision for Asian populations [38,39]. It might be that the differences were not observed owing to the lowering of the threshold for the classification of a patient as obese. At the same time, variables that were possibly collinear were included in the definition of metabolic syndrome. Despite that, pre-transplantation metabolic syndrome is significantly associated with abnormal CA, and the

PANG, KOW, LAW ET AL

main factors contributing to metabolic syndrome were hypertension, dyslipidemia, and DM, all of which were predictors of abnormal CA in our analysis. All factors with significant associations with abnormal CA on univariate analysis were included in the formulation of the predictive model. Although it may be more inconvenient to use numerous factors in a predictive model, the inclusion of more factors increases the accuracy of the model. Careful consideration of these concerns along with the limitations imposed by the sample size led to the incorporation of the 5 variables in the predictive model as described above. There has been no other study that has generated a predictive model for abnormal CA in pre-LT candidates. This model allows for an objective assessment of each individual transplant candidate’s risk for abnormal CA. This provides clinicians with a powerful tool for a much more rigorous selection of patients to undergo CA, dramatically reducing the number of unnecessary CA by up to one-half. This represents a significant proportion of the transplantation population, and the potential benefits of avoiding the risks associated with CA can not be ignored. At the same time, this model has the potential also to identify low-risk patients who may benefit from noninvasive investigations instead. Although the risk factors identified are not new, the regression model gives weight to each factor and provides a clearer representation of how each factor influences the selection process for CA as part of the pre-transplantation work-up. It is important to recognize that the factors do not play the same or equal roles in influencing the severity of the coronary atherosclerosis process in LT recipients. Limitations

It must be recognized that there are several limitations to this study. First, there would have been interobserver variability in the interpretation of CA. Second, because this was a retrospective study, there are inherent limitations in the design of the study and the extent of information available for use. For example, the duration of various chronic conditions, such as DM and hypertension, may not be well documented in all patients and this information therefore not available for analysis. Finally, although the predictive model appears to allow better selection of patients who will benefit from CA and a reduction of unnecessary CA, validation in future prospective studies is required before its use in a clinical setting. This is especially so because the sample size on which the model is based was small. CONCLUSION

This study indicates that DM, dyslipidemia, IHD, age
65 years, and LVH can be used as predictors of abnormal CA in pre-LT patients. The predictive model that has been developed may aid clinicians in better selecting patients who might benefit from CA as part of their pre-transplantation cardiac work-up, although further validation of the model in a large prospective trial is required.

CORONARY ANGIOGRAPHY BEFORE LIVER TRANSPLANT

ACKNOWLEDGMENT The authors thank Ms Koh Wai Ling Hiromi and Associate Professor Tai Bee Choo from the Saw Swee Hock School of Public Health for their invaluable advice on this project.

REFERENCES [1] Carey WD, Dumot JA, Pimentel RR, et al. The prevalence of coronary artery disease in liver transplant candidates over age 50. Transplantation 1995;59:859e64. [2] Plotkin JS, Benitez RM, Kuo PC, et al. Dobutamine stress echocardiography for preoperative cardiac risk stratification in patients undergoing orthotopic liver transplantation. Liver Transpl Surg 1998;4:253e7. [3] Tiukinhoy-Laing SD, Rossi JS, Bayram M, et al. Cardiac hemodynamic and coronary angiographic characteristics of patients being evaluated for liver transplantation. Am J Cardiol 2006;98: 178e81. [4] Johnston SD, Morris SK, Cramb R, Gunson BK, Neuberger J. Cardiovascular morbidity and mortality after orthotopic liver transplantation. Transplantation 2002;73:901e6. [5] Plotkin JS, Scott VL, Pinna A, Dobsch BP, de Wolf AM, Kang Y. Morbidity and mortality in patients with coronary artery disease undergoing orthotopic liver transplantation. Liver Transpl Surg 1996;2:426e30. [6] Diedrich DA, Findlay JY, Harrison BA, Rosen CB. Influence of coronary artery disease on outcomes after liver transplantation. Transplant Proc 2008;40:3554e7. [7] Ali A, Bhardwaj HL, Heuman DM, Jovin IS. Coronary events in patients undergoing orthotopic liver transplantation: perioperative evaluation and management. Clin Transplant 2013;27:e207e15. [8] Murray KF, Carithers Jr RL. AASLD practice guidelines: evaluation of the patient for liver transplantation. Hepatology 2005;41:1407e32. [9] Poldermans D, Bax JJ, Boersma E, et al. Guidelines for preoperative cardiac risk assessment and perioperative cardiac management in noncardiac surgery. Eur Heart J 2009;30:2769e812. [10] Raval Z, Harinstein ME, Skaro AI, et al. Cardiovascular risk assessment of the liver transplant candidate. J Am Coll Cardiol 2011;58:223e31. [11] Fleisher LA, Beckman JA, Brown KA, et al. ACC/AHA 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation 2007;116:e418e500. [12] Lentine KL, Costa SP, Weir MR, et al. Cardiac disease evaluation and management among kidney and liver transplantation candidates. J Am Coll Cardiol 2012;60:434e80. [13] Fleisher LA. Cardiac risk stratification for noncardiac surgery: update from the American College of Cardiology/American Heart Association 2007 guidelines. Cleve Clin J Med 2013;76: S9e15. [14] Hillis LD, Smith PK, Anderson JL, et al. ACCF/AHA guideline for coronary artery bypass graft surgery: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Ann Coll Cardiol 2011;58:e123e210. [15] Grundy SM, Brewer Jr B, Cleeman JI, Smith Jr SC, Lenfant C, NHLBI/AHA conference proceedings. Definition of Metabolic syndrome. Report of the National Heart, Lung, and Blood Institute/American Heart Association Conference on Scientific Issues Related to Definition. Circulation 2004;109:433e8. [16] d’Agostino Sr RB, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care. The Framingham Heart Study. Circulation 2008;117:743. [17] Mells G, Neuberger J. Reducing the risks of cardiovascular disease in liver allograft recipients. Transplantation 2007;83:1141e50.

1805 [18] Guckelberger O, Mutzke F, Glanemann M, et al. Validation of cardiovascular risk scores in a liver transplant population. Liver Transpl 2006;12:394e401. [19] Guckelberger O, Byram A, Klupp J, et al. Coronary event rates in liver transplant recipients reflect the increased prevalence of cardiovascular risk-factors. Transpl Int 2005;18:967e74. [20] Ripoll C, Yotti R, Bermejo J, Bañares R. The heart in liver transplantation. J Hepatol 2011;54:810e22. [21] Keeling AN, Flaherty JD, Davarpanah AH, et al. Coronary multidetector computed tomographic angiography to evaluate coronary artery disease in liver transplant candidates: methods, feasibility and initial experience. J Cardiovasc Med (Hagerstown) 2011;12:460e8. [22] Garg A, Armstrong WF. Echocardiography in liver transplant candidates. JACC Cardiovasc Imaging 2013;6:105e19. [23] Patel S, Kiefer TL, Ahmed A, et al. Comparison of the frequency of coronary artery disease in alcohol-related versus nonalcoholrelated endstage liver disease. Am J Cardiol 2011;108:1552e5. [24] Donovan CL, Marcovitz PA, Punch JD, et al. Two-dimensional and dobutamine stress echocardiography in the preoperative assessment of patients with end-stage liver disease prior to orthotopic liver transplantation. Transplantation 1996;61:1180e8. [25] Harinstein ME, Flaherty JD, Ansari AH, et al. Predictive value of dobutamine stress echocardiography for coronary artery disease detection in liver transplant candidates. Am J Transplant 2008;8:1523e8. [26] Aydinalp A, Bal U, Atar I, et al. Value of stress myocardial perfusion scanning in diagnosis of severe coronary artery disease in liver transplantation candidates. Transplant Proc 2009;41:3757e60. [27] Davidson CJ, Gheorghiade M, Flaherty JD, et al. Predictive value of stress myocardial perfusion imaging in liver transplant candidates. Am J Cardiol 2002;89:359e60. [28] Williams K, Lewis JF, Davis G, Geiser EA. Dobutamine stress echocardiography in patients undergoing liver transplantation evaluation. Transplantation 2000;69:2354e6. [29] Pillarisetti J, Patel P, Duthuluru S, et al. Cardiac catheterization in patients with end-stage liver disease: safety and outcomes. Catheter Cardiovasc Interv 2011;77:45e8. [30] Sharma M, Yong C, Majure D, et al. Safety of cardiac catheterization in patients with end-stage liver disease awaiting liver transplantation. Am J Cardiol 2009;103:742e6. [31] Mandell MS. Cardiovascular disease in liver transplant candidates. Trends Anaesth Crit Care 2011;1:219e23. [32] Targher G, Arcaro G. Nonalcoholic fatty liver disease and increased risk of cardiovascular disease. Atherosclerosis 2007;191: 235e40. [33] London RM, George J. Pathogenesis of NASH: animal models. Clin Liver Dis 2007;22:55e74. [34] Targher G, Bertolini L, Padovani R, Rodella S, Arcaro G, Day C. Differences and similarities in early atherosclerosis between patients with nonalcoholic steatohepatitis and chronic hepatitis B and C. J Hepatol 2007;46:1126e32. [35] Park CW, Tsai NT, Wong LL. Implications of worse renal dysfunction and medical comorbidities in patients with NASH undergoing liver transplant evaluation: impact on MELD and more. Clin Transplant 2011;25:e606e11. [36] Lazo M, Clark JM. The epidemiology of nonalcoholic fatty liver disease: a global perspective. Semin Liver Dis 2008;28:341. [37] Angulo P. Nonalcoholic fatty liver disease and liver transplantation. Liver Transpl 2006;12:253. [38] Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults, National Cholesterol Education Program. Third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation 2002;106:3143e421. [39] World Health Organization. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet 2004;363:157e63.